JP5615252B2 - Appearance inspection device - Google Patents

Description

  The present invention relates to an appearance inspection apparatus that inspects a pattern formed on a surface of a substrate.

  Conventionally, it is known to inspect a pattern formed on the surface of a substrate using an appearance inspection apparatus. In such an appearance inspection apparatus, basically, a sample pattern as a non-defective product is prepared in advance, and the inspection target pattern (the substrate) is compared with the inspection target pattern of the substrate to be inspected. Inspect whether or not is a good product. Here, in the inspection target pattern formed on the substrate, there may be no problem in performance even if the position where the pattern is formed and the color unevenness are different. For such a pattern to be inspected, it is conceivable to lower the detection sensitivity when compared with the sample pattern, but it is difficult to ensure the inspection accuracy.

  For this reason, when the yield rate is lower than a predetermined value, a new sample pattern is created based on the inspection target patterns of a plurality of substrates to be inspected, and the pitch inspection is performed simultaneously. An appearance inspection apparatus that sets detection sensitivity so as to enable detection of the same defect in inspection target patterns of a plurality of substrates to be inspected based on comparison with the results of pitch inspection has been considered (for example, Patent Document 1). reference). With this appearance inspection apparatus, it is possible to prevent a reduction in the yield rate while ensuring the work efficiency of the inspection.

JP 2010-117132 A

  However, in the pattern to be inspected, even in a range that does not cause a problem in performance, the formed position and color unevenness are different, and there is a case where a certain tendency is not seen (there is variation) in the difference. . For this reason, in the above-described conventional visual inspection apparatus, it is difficult to ensure inspection accuracy for such an inspection target pattern.

  The present invention has been made in view of the above circumstances, and its purpose is that the position and color unevenness to be formed are different within a range that does not cause a problem in performance, and a certain tendency is seen in the difference. An object of the present invention is to provide an appearance inspection apparatus that can ensure inspection accuracy even for a pattern to be inspected (which has variations).

  According to the first aspect of the present invention, an inspection mechanism that acquires an image of an inspection target pattern of an inspection target, and image data acquired by the observation mechanism are analyzed using a sample pattern prepared in advance, and the quality of the inspection target is determined. And an image processing unit that determines the boundary line in the sample pattern for each of a plurality of partition regions along the boundary line of the sample pattern as an analysis process. An average boundary image that is an average value of the boundary lines of the inspection target pattern viewed in the orthogonal boundary coordinate directions is generated, and it is determined whether or not the matching ratio between the average boundary image and the sample pattern is within an allowable range. And an item for extracting a defect candidate by obtaining a difference between the reference boundary image generated from the average boundary image and the inspection target pattern, and determining whether the defect candidate is within an allowable range; The Appearance inspection apparatus and performs a boundary test process of said object pattern row.

  A second aspect of the present invention is the appearance inspection apparatus according to the first aspect, wherein the image processing unit includes each pixel in a search area including the sample pattern and set to an area larger than the sample pattern. A pixel close to the color centroid of the sample pattern is extracted, and a fixed region set as a region smaller than the sample pattern is applied to the inside of the sample pattern to the color-extracted pixel group. On the other hand, an extraction pattern is generated by applying image data as the sample pattern at a location corresponding to the determined region, and the extraction pattern is used for the analysis process of the inspection target pattern.

  A third aspect of the present invention is the appearance inspection apparatus according to the second aspect, wherein the image processing unit determines whether or not the total area of the extracted patterns is within an allowable range as an analysis process. An item for determining whether the perimeter of the extraction pattern is within an allowable range, an item for determining whether the vertical dimension of the extraction pattern is within an allowable range, An item for determining whether or not a dimension is within an allowable range, and an item for determining whether or not a correlation value indicating a degree of difference between the extracted pattern and the sample pattern is within an allowable range. It is characterized in that a shape inspection process of the inspection target pattern is performed.

  A fourth aspect of the present invention is the appearance inspection apparatus according to the second or third aspect, wherein the image processing unit includes, as an analysis process, a deviation of a luminance histogram of the extracted pattern within an allowable range. Whether or not the roughness evaluation value calculated by differentiating the image data of the extracted pattern and the item for determining whether or not there is, and dividing the sum of the differential values by the number of pixels of the extracted pattern is within an allowable range And a color inspection process for the inspection target pattern to be executed.

  According to the appearance inspection apparatus of the present invention, it is determined that there is no problem in the boundary inspection unless there is a large deviation and there is no large unevenness (defect) when viewed in units of divided areas. Is judged to be a non-defective product even if it has a different shape, a boundary line (outer) is formed at a position different from the sample pattern, or irregularities (defects) that are not in the sample pattern are formed. be able to.

  In addition, by appropriately setting each allowable range, it is possible to appropriately set the size of the problem in the deviation of the boundary line (outside) with respect to the sample pattern and the size of the problem in the unevenness (defect) that does not exist in the sample pattern. If there is a large shift or a large unevenness (defect) when viewed in units of divided areas, it is determined that there is a problem in the boundary inspection, so that the inspection accuracy can be ensured.

  For this reason, within the range that does not pose a problem in performance, the position to be formed, color unevenness, etc. are different, and even if there is a certain tendency (difference) in the difference, inspection accuracy Can be secured.

  In addition to the above-described configuration, the image processing unit extracts pixels close to the color centroid of the sample pattern from among the pixels in the search area that includes the sample pattern and is set to an area larger than the sample pattern. Then, a fixed region set as a region smaller than the sample pattern is applied to the inside of the sample pattern to the color-extracted pixel group, and the sample pattern of the portion corresponding to the fixed region is located inside the fixed region. When an extraction pattern is generated by applying image data, and the extraction pattern is used for the analysis process of the inspection target pattern, each inspection ( Shape, color, and boundary), and it is possible to generate an extraction pattern that represents only the outer shape necessary for inspection, making it simpler And it can be appropriately each test (shape, color and border).

  In addition to the configuration described above, the image processing unit, as an analysis process, determines whether or not the total area of the extracted pattern is within an allowable range, and the perimeter of the extracted pattern is within the allowable range. An item for determining whether the vertical dimension of the extraction pattern is within an allowable range, an item for determining whether the horizontal dimension of the extraction pattern is within an allowable range, If the shape inspection process of the inspection target pattern is executed, the item for determining whether or not the correlation value indicating the degree of difference between the extracted pattern and the sample pattern is within an allowable range is somewhat similar. If there is no problem, the shape is different from the sample pattern, is different in size from the sample pattern, is formed at a different position from the sample pattern, Even when they happen to be formed over not in the emission irregularities (defects), it can be determined as a good product.

  In addition to the above-described configuration, the image processing unit performs, as analysis processing, an item for determining whether or not the deviation of the histogram of the luminance of the extracted pattern is within an allowable range, and differential processing of the image data of the extracted pattern And determining whether or not the rough evaluation value calculated by dividing the sum of the differential values by the number of pixels of the extracted pattern is within an allowable range. If processing is performed, it is judged that there is no problem if there is no part where the color tone is similar to the whole and the color changes extremely, so the color is different from the sample pattern or the sample pattern Even if it is a different color scheme, it can be determined as a non-defective product.

It is explanatory drawing which shows typically the structure of the external appearance inspection apparatus 10 as an example of the inspection apparatus which concerns on this invention. It is explanatory drawing which shows typically the wafer 15 as a test object. It is explanatory drawing which shows typically an example of the semiconductor element 16 (its test object pattern Po) as a test object. FIG. 3 is an explanatory diagram schematically showing a sample pattern Pm (its image data) in Example 1. It is explanatory drawing which shows a mode that the search area | region Ar and the determination area | region Af are produced from the sample pattern Pm. 6 is a flowchart illustrating an example of the content of an inspection target pattern Po extraction process in the image processing unit 32; It is explanatory drawing which shows a mode that the search area | region Ar was set to the semiconductor element 16 (the image data) in an observation position. It is explanatory drawing which shows a mode that the data of the sample pattern Pm in the definite area | region Af were superimposed on the pixel group (inspection target pattern Po) which carried out color extraction. 6 is a flowchart illustrating an example of the content of a shape inspection process of an inspection target pattern Po in an image processing unit 32. It is a flowchart which shows an example of the content of the color inspection process of the test target pattern Po in the image process part 32. FIG. It is a graph which shows the deviation Hd of a histogram, The value (lightness) in the image data of the extraction pattern Pa is shown on a horizontal axis, and the frequency (frequency) in each Value (lightness) is shown on the vertical axis | shaft. 5 is a flowchart illustrating an example of the content of a boundary inspection process for an inspection target pattern Po in an image processing unit 32. It is explanatory drawing which shows the boundary coordinate direction Y in the division area Dn, and the mode of a scan, (a) shows the mode in the division area D2, (b) has shown the mode in the division area D5. FIG. 6 is an explanatory diagram showing a boundary line (outer shape) of a sample pattern Pm as an origin in a boundary coordinate direction Y in order to explain a state in which an average boundary image Ib of an extraction pattern Pa is generated using a partition area D2 as an example; Is schematically shown by superimposing the boundary line of the sample pattern Pm on the boundary line (its image data) of the extracted pattern Pa, and (b) schematically shows the average boundary image Ib. FIG. 6 is an explanatory diagram showing a reference boundary image Bb generated from an average boundary image Ib with the boundary line (outer contour) of the sample pattern Pm as the origin in the boundary coordinate direction Y. It is explanatory drawing which shows a mode that the difference between reference | standard boundary image Bb (the image data) and extraction pattern Pa (the image data) is calculated | required, and defect candidate Lc is extracted. It is explanatory drawing which shows the example where the deviation Hd of the histogram of the brightness | luminance of the extraction pattern Pa as a whole is mutually equal, and the roughness evaluation value Vr differs, (a) is extraction pattern Pa (inspection target pattern Po) gradation form. (B) shows a case where the extraction pattern Pa (inspection target pattern Po) is patchy.

  Embodiments of an appearance inspection apparatus according to the present invention will be described below with reference to the drawings.

  First, a schematic configuration of an appearance inspection apparatus 10 according to an embodiment of the appearance inspection apparatus according to the present invention will be described. FIG. 1 is an explanatory view schematically showing a configuration of an appearance inspection apparatus 10 as an example of an inspection apparatus according to the present invention. FIG. 2 is an explanatory diagram schematically showing a wafer 15 as an inspection target. FIG. 3 is an explanatory diagram schematically showing an example of the semiconductor element 16 (its inspection target pattern Po) as an inspection target.

  As shown in FIG. 1, the appearance inspection apparatus 10 is generally configured by an observation mechanism 11, an illumination mechanism 12, an operation holding mechanism 13, and a control mechanism 14. This appearance inspection apparatus 10 basically holds a wafer 15 to be inspected by an operation holding mechanism 13 and observes the wafer 15 by an observation mechanism 11 while illuminating the wafer 15 by an illumination mechanism 12.

  On the wafer 15, a plurality of semiconductor elements 16 are formed in alignment as shown in FIG. As shown in FIG. 3, each of the semiconductor elements 16 is configured by forming a pattern (inspection target pattern Po) with an exposure apparatus (not shown). In each semiconductor element 16 of the present embodiment, two alignment marks Am are provided outside the pattern (inspection target pattern Po). In order to inspect whether or not the pattern in each semiconductor element 16 is appropriately formed, the appearance inspection apparatus 10 (see FIG. 1) is used. For this reason, the pattern in each semiconductor element 16 becomes the inspection object pattern Po as the inspection object.

  As shown in FIG. 1, the observation mechanism 11 of the appearance inspection apparatus 10 includes an objective lens barrel 21 and an imaging camera 22 attached thereto. Although not shown, the objective lens barrel 21 is provided with a relay lens and a revolver type turret, and a plurality of objective lenses can be switched and set by the turret. An imaging camera 22 is attached to the upper part of the objective lens barrel 21. The imaging camera 22 passes through the objective lens barrel 21 (objective lens and relay lens) according to the magnification determined by the objective lens set in the objective lens barrel 21, and on the optical axis (observation optical axis Oa). Image data at a predetermined position can be acquired. For this reason, the imaging camera 22 and the objective lens barrel 21 function as an observation optical system in the observation mechanism 11, and the optical axis thereof becomes the observation optical axis Oa. A plane (object-side imaging plane) that includes the position at which the imaging camera 22 can acquire an appropriate image on the observation optical axis Oa, that is, the focal position in the observation optical system and is orthogonal to the observation optical axis Oa is the observation plane Fp Become. In the following, the direction of the observation optical axis Oa is the Z-axis direction, and the plane orthogonal to the direction is the XY plane. The image data acquired by the imaging camera 22 can be displayed on the display unit 34 by the control mechanism 14 (the operation control unit 26 described later) and is appropriately analyzed by the image processing unit 32 described later.

  In the appearance inspection apparatus 10, an illumination mechanism 12 is provided to appropriately observe the inspection object (wafer 15) by the observation mechanism 11. The illumination mechanism 12 has an illumination unit 23. The illumination unit 23 has a light emitting unit (not shown), and the light emitting unit is appropriately controlled to be turned on and off under the control of an illumination driving unit 28 described later. The illumination unit 23 can illuminate the observation surface Fp with the irradiation light emitted from the light emitting unit in a direction inclined with respect to the observation optical axis Oa.

  An operation holding mechanism 13 is provided below the objective lens barrel 21 and the illumination unit 23. The operation holding mechanism 13 includes a holding stage 24 and an operation unit 25. The holding stage 24 has a disk shape as a whole, and has an upper end surface as a mounting surface 24a that exists along a plane orthogonal to the observation optical axis Oa. The holding stage 24 can hold the inspection target (wafer 15 in this embodiment) placed on the placement surface 24a under the control of an operation holding drive unit 29 described later. The holding on the holding stage 24 is omitted if it is fixed to the mounting surface 24a without causing a positional shift when the inspection object (wafer 15) for alignment on the mounting surface 24a is moved. May be adsorbed onto the mounting surface 24a, or may be pressed against the mounting surface 24a with a claw that can be moved forward and backward. In the present embodiment, the holding stage 24 holds the inspection object (wafer 15) by being attracted to the mounting surface 24a by driving a suction device (not shown). Thereby, the inspection object (wafer 15) is arranged on the observation surface Fp. The holding stage 24 is provided on the operating unit 25.

  The operating unit 25 can move the holding stage 24 (the mounting surface 24a) in the X-axis direction, move it in the Y-axis direction, and move it in the Z-axis direction. ing. In addition, the operating unit 25 can rotate the holding stage 24 (its mounting surface 24a) about the Z axis and can adjust the inclination of the mounting surface 24a with respect to the Z axis direction. . The operation unit 25 can appropriately move the holding stage 24 (the mounting surface 24a) under the control of an operation holding drive unit 29 described later.

  For this reason, the holding stage 24 drives the operation unit 25 under the control of an operation holding drive unit 29 described later, whereby the inspection object (wafer 15) held on the mounting surface 24a is placed on the observation optical axis Oa. It can be moved along a plane that is orthogonal, can be rotated on a plane that is orthogonal to the observation optical axis Oa, and the inclination with respect to the observation optical axis Oa can be adjusted. Thus, the operation holding mechanism 13 can adjust (alignment) the position and orientation of the inspection target (wafer 15) with respect to the observation optical axis Oa.

  In the appearance inspection apparatus 10, the objective lens barrel 21 and the imaging camera 22 of the observation mechanism 11 configured as described above, the illumination unit 23 of the illumination mechanism 12, the holding stage 24 and the operation unit 25 of the operation holding mechanism 13. Are controlled under the control of the control mechanism 14. The control mechanism 14 includes an arithmetic control unit 26, an observation drive unit 27, an illumination drive unit 28, an operation holding drive unit 29, a storage unit 31, and an image processing unit 32.

  The arithmetic control unit 26 comprehensively controls the control in the control mechanism 14, that is, the operation of the appearance inspection apparatus 10. The arithmetic control unit 26 appropriately drives each unit of the appearance inspection apparatus 10 based on an operation performed on the operation unit 33 and causes the display unit 34 to appropriately display a video based on the image data acquired by the imaging camera 22. In addition, the arithmetic control unit 26 can store the image data output from the observation driving unit 27 in the storage unit 31 and appropriately extract the image data from the storage unit 31.

  The observation drive unit 27 controls operations such as focus adjustment and magnification conversion in the objective lens barrel 21 and controls operations such as acquisition of image data and output of the acquired image data in the imaging camera 22. For this reason, the observation drive unit 27 constitutes the observation mechanism 11 in cooperation with the objective lens barrel 21 and the imaging camera 22. The observation mechanism 11 can acquire an image on the observation surface Fp that has passed through the objective lens barrel 21 with the imaging camera 22 under the control of the observation drive unit 27.

  The illumination driving unit 28 appropriately controls turning on / off of the illumination unit 23 (its light emitting unit (not shown)). The illumination drive unit 28 can irradiate the observation surface Fp with observation illumination or inspection illumination by appropriately turning on / off the illumination unit 23. For this reason, the illumination driving unit 28 constitutes the illumination mechanism 12 in cooperation with the illumination unit 23. The illumination mechanism 12 can assist the acquisition of an image on the observation surface Fp by the observation mechanism 11 under the control of the illumination drive unit 28.

  The operation holding drive unit 29 controls driving of the holding stage 24 and the operation unit 25 as appropriate. That is, the operation holding drive unit 29 controls the suction device (not shown) of the holding stage 24 as appropriate so that the holding stage 24 sucks and holds the inspection target (wafer 15 in this embodiment), and Can be switched. Further, the operation holding drive unit 29 appropriately controls the operation unit 25 to control the holding stage 24, that is, the inspection object (wafer 15) sucked and held on the observation optical axis Oa (observation optical system). In addition, it can be appropriately moved on the XY plane orthogonal thereto, can be appropriately rotated around the observation optical axis Oa, and the inclination with respect to the observation optical axis Oa can be adjusted. As a result, it is possible to inspect an inspection object (semiconductor element 16) at an arbitrary position on the inspection object (wafer 15) held by suction by the holding stage 24. Therefore, the operation holding drive unit 29 constitutes the operation holding mechanism 13 in cooperation with the holding stage 24 and the operation unit 25. The operation holding mechanism 13 can appropriately acquire an image on the observation surface Fp by the observation mechanism 11 under the control of the operation holding drive unit 29.

  The image processing unit 32 appropriately analyzes the image data acquired by the imaging camera 22 (observation mechanism 11). This analysis includes comparison of image data for inspection, recognition of outlines of a plurality of inspection objects (semiconductor elements 16), determination of a focus state at an observation point, Recognizing defects, recognizing defects such as electrodes and wiring (pattern) provided on each inspection object, and recognizing defective cutting of each inspection object (semiconductor element 16) from the wafer 15 Etc. The determination as to whether or not the inspection target pattern Po in the analysis processing of the image processing unit 32 is appropriately formed will be described later in detail.

  In the appearance inspection apparatus 10 that is comprehensively controlled under the control of the control mechanism 14 (arithmetic control unit 26), all inspection objects (semiconductor elements 16) in the inspection object (wafer 15) are inspected as follows. Do.

  In the appearance inspection apparatus 10, first, a suction device (not shown) is driven by the operation holding drive unit 29, whereby the operation holding mechanism 13 sucks and holds the wafer 15 to be inspected and performs chip scanning. In this chip scan, a surface (XY) perpendicular to the observation optical axis Oa is operated by the operation and holding drive unit 29 so as to scan the mounting surface 24a of the holding stage 24 in the operation and holding mechanism 13, that is, the wafer 15. The holding stage 24 is moved along the plane. In this scanning, an image is acquired by the imaging camera 22 while being illuminated by the illumination mechanism 12.

  Next, a chip map is created. This chip map indicates the position and orientation of each inspection object (semiconductor element 16) on the wafer 15. The control mechanism 14 discriminates a plurality of inspection objects (semiconductor elements 16) by recognizing the contour line by chip scanning, and takes into consideration the movement position information by the operation holding drive unit 29, whereby each inspection object. The position and orientation of (semiconductor element 16) are acquired.

  Next, the inspection object (semiconductor element 16) is determined. In this determination, an observation position by the observation optical system is determined based on the created chip map, and the holding stage 24 is moved by the operation holding drive unit 29 according to the determination. Thereafter, the quality of the inspection object (semiconductor element 16 (its inspection object pattern Po)) at the observation position is determined by analysis processing in the image processing unit 32 as described later.

  The inspection of the wafer 15 is completed by sequentially determining all of the inspection objects (semiconductor elements 16) on the wafer 15 based on the created chip map. To do. This inspection can also be performed by visually observing an image based on the image data acquired by the imaging camera 22 displayed on the display unit 34 under the control of the arithmetic control unit 26.

  Next, analysis processing for determining the quality of the inspection object (semiconductor element 16) at the observation position in the image processing unit 32 will be described with reference to FIGS. The image processing unit 32 appropriately analyzes the image data of the inspection target pattern Po of the semiconductor element 16 as the inspection target acquired by the imaging camera 22 to determine whether or not the inspection target pattern Po is appropriately formed. To determine whether the semiconductor element 16 (inspection object) is good or bad. Even if the inspection target pattern Po is in a range that does not cause a problem in performance, the position to be formed, color unevenness, and the like are different, and there is a case where a certain tendency is not seen in the difference (variation). . For this reason, in the analysis processing in the image processing unit 32 of the present invention, it is appropriately determined whether or not the inspection target pattern Po is appropriately formed while dealing with such variations in a range that does not cause a problem in performance. It is possible to do. At this time, the image processing unit 32 basically determines the quality of the semiconductor element 16 by comparison with a sample pattern Pm (see FIG. 4) which is the inspection target pattern Po as a non-defective product.

  The sample pattern Pm shown in FIG. 4 is prepared in advance corresponding to the semiconductor element 16 to be inspected and stored in the storage unit 31 (see FIG. 1). In this embodiment, the sample pattern Pm is created and stored as image data from an average value of a plurality of non-defective inspection target patterns Po of the semiconductor element 16. As data of this sample pattern Pm, as will be described later, the position, shape, color centroid, total area, and perimeter of the semiconductor element 16 are used. In this embodiment, the position in the semiconductor element 16 is set with reference to two alignment marks Am. Since the sample pattern Pm is stored as image data in this embodiment, the position and shape can be acquired from the position of each pixel constituting the sample pattern Pm (its image data). Further, the color centroid can be acquired from the color reproduced by each pixel constituting the sample pattern Pm (its image data). Further, the total area can be obtained from the number of all pixels constituting the sample pattern Pm (its image data). Next, the perimeter can be obtained from the number of pixels constituting the boundary line (outline) of the sample pattern Pm (its image data).

  In the sample pattern Pm, n partition regions Dn (n is a natural number excluding 0) are set along the boundary (outer). Each partition region Dn divides a boundary line (outer shape) in the sample pattern Pm into n pieces. The dividing method as each partition region Dn uses a point where the property of the boundary line (outer shape) in the sample pattern Pm changes as a boundary. In other words, a portion where the same pattern is continuously formed on the boundary line of the sample pattern Pm is defined as one partition region Dn. The property of this boundary line (outer shape) refers to the shape (whether it is linear or curved), the extending direction, the color to be formed, and the like. In this embodiment, seven partitioned areas D1 to D7 are set for the sample pattern Pm.

  As shown in FIG. 5, the image processing unit 32 creates a search area Ar and a confirmed area Af from the sample pattern Pm. The search area Ar defines an area in which the search for the inspection target pattern Po of the semiconductor element 16 that is the inspection target is executed. The search area Ar is set as an area that includes the sample pattern Pm and is larger than the sample pattern Pm. In this embodiment, the sample pattern Pm is expanded at a certain rate (for example, the outline of the sample pattern Pm). Is expanded to the outside of 7 pixels).

  The fixed region Af defines a region where the inspection target pattern Po is formed in any case where variations occur in a range that does not cause a problem in performance in the inspection target pattern Po. This fixed region Af is set as a region smaller than the sample pattern Pm inward of the sample pattern Pm. In this embodiment, the sample pattern Pm is contracted at a certain rate (for example, the sample pattern Pm The outline is reduced to the inside of 5 pixels). Note that the amount of expansion in the search area Ar and the amount of contraction in the fixed area Af can be changed as appropriate. For this reason, the expansion amount and the contraction amount can be appropriately set according to the variation in the pattern (inspection target pattern Po) of the inspection target (semiconductor element 16 in this embodiment).

  Next, the extraction process of the inspection target pattern Po in the analysis process of the image processing unit 32 will be described. The image processing unit 32 first extracts the inspection target pattern Po of the semiconductor element 16 in order to analyze the image data of the inspection target pattern Po of the semiconductor element 16 that is the inspection target. FIG. 6 is a flowchart illustrating an example of the contents of the inspection target pattern Po extraction process in the image processing unit 32. FIG. 7 is an explanatory diagram showing a state where the search area Ar is set in the semiconductor element 16 (its image data) at the observation position. FIG. 8 is an explanatory diagram showing a state in which the data of the sample pattern Pm in the defined area Af is superimposed on the color-extracted pixel group (inspection target pattern Po). Hereinafter, each step of the flowchart of FIG. 6 which is an example of the control process in the image processing unit 32 in the extraction process will be described with reference to FIGS.

  In the flowchart of FIG. 6, it is assumed that the semiconductor element 16 (its inspection target pattern Po) shown in FIG. 3 is acquired by the imaging camera 22 as the semiconductor element 16 (its image data) at the observation position.

  In step S1, the search area Ar is set in the acquired image of the semiconductor element 16, and the process proceeds to step S2. In this step S1, as shown in FIG. 7, by superimposing the data of the search area Ar on the acquired image data of the semiconductor element 16, a search area Ar for searching the inspection target pattern Po in the semiconductor element 16 is formed. Set. At this time, the data in the search region Ar is superimposed on the two alignment marks Am in the obtained semiconductor element 16 in a positional relationship where the two alignment marks Am in the sample pattern Pm match.

  In step S2, following the setting of the search area Ar in step S1, color extraction in the search area Ar is performed, and the process proceeds to step S3. In this step S2, a pixel close to the color centroid in the sample pattern Pm is extracted from each pixel in the search area Ar. The term “close to the color center of gravity” as used herein means that each color constituting the color image is within a predetermined value range from the numerical value representing the sample pattern Pm. The predetermined value can be set as appropriate. For this reason, the color extraction range can be appropriately set according to the variation in the pattern (inspection target pattern Po) of the inspection target (semiconductor element 16 in this embodiment). Further, each color constituting the color image means HSV (Hue (Hue), Saturation (Saturation), Value (Lightness)), RGB (Red (Red), Green (Green), Blue (Blue)) or the like. . In this embodiment, pixels within a predetermined value range are extracted from the HSV value representing the sample pattern Pm.

  In step S3, following the color extraction in the search area Ar in step S2, the determined area Af is applied to the color-extracted pixel group, and the process proceeds to step S4. In step S3, as shown in FIG. 8, the data of the definite region Af is superimposed on the pixel group (see reference symbol Pa (Po)) extracted in step S2. As a result, in the pixel group extracted in step S2, when a hollow portion E is generated at the center position (see FIGS. 3 and 7), the hollow portion E exists in the fixed region Af. .

  In step S4, following the application of the fixed region Af to the color-extracted pixel group in step S3, an extraction pattern Pa (its image data) is generated, and this flowchart is ended. In this step S4, the image data as the sample pattern Pm of the portion corresponding to the fixed region Af is placed inward of the fixed region Af applied in step S3 in the pixel group (image data by them) extracted in step S2. The extraction pattern Pa is generated by applying (substituting the image data of the sample pattern Pm corresponding to the inside of the fixed region Af). This extraction pattern Pa is used for comparison with the sample pattern Pm for inspection of the inspection target pattern Po. As a result, in the pixel group extracted in step S2, even if a hollow portion E is generated as shown in FIGS. 3 and 7, the extracted pattern Pa is shown in FIG. The missing portion E is filled (the hollow portion E is filled). Thereby, extraction pattern Pa (its image data) as inspection object pattern Po of semiconductor element 16 (inspection object) in an observation position can be generated, and extraction processing of inspection object pattern Po is completed.

  Thereafter, the image processing unit 32 performs a shape inspection process of the inspection target pattern Po in the analysis process. In this shape inspection process, the shape inspection of the inspection target pattern Po is performed by comparing the extracted pattern Pa (its image data) with the sample pattern Pm. FIG. 9 is a flowchart illustrating an example of the contents of the shape inspection process of the inspection target pattern Po in the image processing unit 32. Hereinafter, each step of the flowchart of FIG. 9 which is an example of the control process in the image processing unit 32 in the shape inspection process will be described.

  In step S11, the total area Sa of the extraction pattern Pa is calculated, and the process proceeds to step S12. In step S11, the total area Sa of the extraction pattern Pa is calculated from the image data as the extraction pattern Pa generated in the flowchart of FIG. In the present embodiment, the total area Sa of the extraction pattern Pa is calculated from the number of all pixels constituting the image data of the extraction pattern Pa.

  In step S12, following the calculation of the total area Sa of the extraction pattern Pa in step S11, it is determined whether or not the total area Sa is within the allowable range. If Yes, the process proceeds to step S13. Proceed to step S22. In step S12, if the total area Sa does not exceed the allowable value Ts with respect to the total area Sm of the sample pattern Pm, it is determined as Yes (Sm−Ts <Sa <Sm + Ts), and the value exceeds the total value. If it is, it will be judged No. Step S12 is an inspection item for inspecting the extracted pattern Pa from the viewpoint of the total area Sa in the shape inspection. This allowable value Ts may be set as a predetermined value, or may be set as a ratio with respect to the total area Sm. This allowable value Ts can be changed as appropriate. For this reason, the difference of the total area Sa with respect to the total area Sm can be appropriately set according to the variation in the pattern (inspection target pattern Po) of the inspection target (semiconductor element 16 in this embodiment).

  In step S13, following the determination that the total area Sa is within the allowable range in step S12, the peripheral length La of the extraction pattern Pa is calculated, and the process proceeds to step S14. In step S13, the perimeter length La of the extraction pattern Pa is calculated from the generated image data as the extraction pattern Pa. In the present embodiment, the perimeter length La of the extraction pattern Pa is calculated from the number of pixels constituting the boundary line (outline) of the extraction pattern Pa (its image data).

  In step S14, following the calculation of the perimeter length La of the extraction pattern Pa in step S12, it is determined whether or not the perimeter length La is within the allowable range. If Yes, the process proceeds to step S15. Proceed to step S22. In step S14, if the peripheral length La does not exceed the permissible value Tl with respect to the peripheral length Lm of the sample pattern Pm, it is determined as Yes (Lm−Tl <La <Lm + Tl), and the value is exceeded. If it is, it will be judged No. Step S14 is an inspection item for inspecting the extracted pattern Pa from the viewpoint of the peripheral length Lm in the shape inspection. This allowable value Tl may be set as a predetermined value or may be set as a ratio with respect to the peripheral length Lm. This allowable value Tl can be changed as appropriate. For this reason, the difference of the peripheral length Lm with respect to the peripheral length Lm can be appropriately set according to the variation in the pattern (inspection target pattern Po) of the inspection target (semiconductor element 16 in this embodiment).

  In step S15, following the determination that the peripheral length La is within the allowable range in step S14, the vertical dimension Mla of the extraction pattern Pa is calculated, and the process proceeds to step S16. In step S15, the vertical dimension Mla of the extraction pattern Pa (the vertical dimension when viewed from the front in FIG. 8) is calculated from the generated image data as the extraction pattern Pa. In this embodiment, among the pixels constituting the extraction pattern Pa (its image data), the pixel at one end (uppermost) and the pixel at the other end (lowermost) as viewed in the vertical direction From the interval, the vertical dimension Mla of the extraction pattern Pa is calculated.

  In step S16, following the calculation of the vertical dimension Mla of the extraction pattern Pa in step S15, it is determined whether or not the vertical dimension Mla is within the allowable range. If Yes, the process proceeds to step S17. Proceed to step S22. In this step S16, if the vertical dimension Mla does not exceed the allowable value Tml with respect to the vertical dimension Mlm of the sample pattern Pm (Mlm−Tml <Mla <Mlm + Tml), it is determined as Yes and becomes the exceeded value. If it is, it will be judged No. This allowable value Tml may be set as a predetermined value or may be set as a ratio with respect to the vertical dimension Mlm. Step S16 is an inspection item for inspecting the extracted pattern Pa from the viewpoint of the vertical dimension Mlm in the shape inspection. This allowable value Tml can be changed as appropriate. For this reason, the difference of the vertical dimension Mla with respect to the vertical dimension Mlm can be appropriately set according to the variation in the pattern (inspection target pattern Po) of the inspection target (semiconductor element 16 in this embodiment).

  In step S17, following the determination in step S16 that the vertical dimension Mla is within the allowable range, the horizontal dimension Mwa of the extraction pattern Pa is calculated, and the process proceeds to step S18. In step S17, the horizontal dimension Mwa of the extraction pattern Pa (the horizontal dimension when viewed from the front in FIG. 8) is calculated from the generated image data as the extraction pattern Pa. In the present embodiment, among each pixel constituting the extraction pattern Pa (its image data), a pixel at one end (leftmost) and a pixel at the other end (rightmost) as viewed in the horizontal direction From the interval, the horizontal dimension Mwa of the extraction pattern Pa is calculated.

  In step S18, following the calculation of the horizontal dimension Mwa of the extraction pattern Pa in step S17, it is determined whether or not the horizontal dimension Mwa is within the allowable range. If Yes, the process proceeds to step S19, and if No, Proceed to step S22. In this step S18, if the horizontal dimension Mwa does not exceed the allowable value Tmw with respect to the horizontal dimension Mwm of the sample pattern Pm (Mwm−Tmw <Mwa <Mwm + Tmw), it is determined as Yes and exceeds the value. If it is, it will be judged No. Step S18 is an inspection item for inspecting the extracted pattern Pa from the viewpoint of the lateral dimension Mwm in the shape inspection. This allowable value Tmw may be set as a predetermined value, or may be set as a ratio to the horizontal dimension Mwm. This allowable value Tmw can be changed as appropriate. For this reason, the difference of the horizontal dimension Mwa with respect to the horizontal dimension Mwm can be appropriately set according to the variation in the pattern (inspection target pattern Po) of the inspection target (semiconductor element 16 in this embodiment).

  In step S19, following the determination that the horizontal dimension Mwa is within the allowable range in step S18, the correlation value C of the extraction pattern Pa is calculated, and the process proceeds to step S20. In this step S19, a correlation value C indicating the degree of difference between the sample pattern Pm and the extracted pattern Pa is calculated with the case where the shape of the sample pattern Pm and the shape of the generated extracted pattern Pa completely match each other as 100%. In this embodiment, the correlation value C forms image data of the sample pattern Pm having a different number of pixels Pd in each pixel forming the image data of the sample pattern Pm and each pixel forming the image data of the extraction pattern Pa. Calculation is performed by obtaining a ratio (C = [Pd / Pw] × 100) to the total number of pixels Pw.

  In step S20, following the calculation of the correlation value C of the extraction pattern Pa in step S19, it is determined whether or not the correlation value C is within the allowable range. If Yes, the process proceeds to step S21. Proceed to step S22. In step S20, if the correlation value C does not exceed the allowable value Tc, it is determined as Yes (C <Tc), and if it exceeds the value, it is determined No. Step S20 is an inspection item for inspecting the extracted pattern Pa from the viewpoint of the correlation value C in the shape inspection. This allowable value Tc is set as a predetermined value. This allowable value Tc can be changed as appropriate. Therefore, the difference of each pixel forming the image data of the extraction pattern Pa with respect to each pixel forming the image data of the sample pattern Pm is determined in the pattern (inspection target pattern Po) of the inspection target (semiconductor element 16 in this embodiment). It can be set as appropriate according to the variation.

  In step S21, following the determination that the correlation value C is within the allowable range in step S20, all the shape inspections of the inspection target pattern Po are completed, and the flowchart is ended assuming that the criteria of each item are satisfied. . In this step S21, there is no problem in the shape inspection of the inspection target pattern Po of the semiconductor element 16 as the inspection target at the observation position, and the shape inspection is performed in order to perform the color inspection processing of the inspection target pattern Po in the analysis processing. The inspection process is terminated. At this time, the fact that there was no problem in the shape inspection is stored in combination with the semiconductor element 16 to be inspected.

  In step S22, it is determined that the total area Sa is not within the allowable range in step S12, or the peripheral length La is not within the allowable range in step S14, or the vertical dimension Mla in step S16 is Following the determination that it is not within the allowable range, the determination that the lateral dimension Mwa is not within the allowable range in step S18, or the determination that the correlation value C is not within the allowable range in step S20, It is assumed that the shape inspection of the target pattern Po has not been completed because the criteria of any item are not satisfied, and this flowchart is ended. In this step S22, the shape inspection process is terminated assuming that there is a problem in the shape inspection of the inspection target pattern Po of the semiconductor element 16 as the inspection target at the observation position. In this embodiment, when the shape inspection process (flowchart in FIG. 9) is completed through step S22, the inspection target pattern Po of the semiconductor element 16 as the inspection target at the observation position is problematic. The analysis process for the semiconductor element 16 is terminated without performing the inspection. At this time, the fact that there is a problem in any item is stored in combination with the semiconductor element 16 to be inspected. Thereby, the shape inspection of the extracted pattern Pa (its image data) as the inspection target pattern Po of the semiconductor element 16 (inspection target) at the observation position is completed.

  Thereafter, the image processing unit 32 performs a color inspection process for the inspection target pattern Po in the analysis process. In this color inspection process, the uniformity of color in the extraction pattern Pa (its image data) described above is determined. FIG. 10 is a flowchart illustrating an example of the content of the color inspection process of the inspection target pattern Po in the image processing unit 32. FIG. 11 is a graph showing the deviation Hd of the histogram, in which the value (frequency) in the image data of the extracted pattern Pa is shown on the horizontal axis, and the frequency (frequency) at each value (value) is shown on the vertical axis. Hereinafter, each step of the flowchart of FIG. 10 which is an example of the control process in the image processing unit 32 in the color inspection process will be described with reference to FIG.

  In step S31, the deviation Hd of the luminance histogram of the extraction pattern Pa generated in the flowchart of FIG. 6 is calculated, and the process proceeds to step S32. In step S31, a histogram indicating the luminance distribution of each pixel in the image data of the extracted pattern Pa generated as the inspection target pattern Po of the semiconductor element 16 at the observation position is generated. Thereafter, the deviation Hd of the generated histogram is calculated. In this embodiment, in step S31, as shown in FIG. 11, the deviation of the histogram with the horizontal axis representing the value (lightness) in the image data of the extracted pattern Pa and the vertical axis representing the frequency (frequency) at each value (lightness). Hd is calculated.

  In step S32, following the calculation of the deviation Hd of the luminance histogram of the extraction pattern Pa in step S31, it is determined whether or not the deviation Hd is within the allowable range. If Yes, the process proceeds to step S33. Advances to step S36. In this step S32, if the deviation Hd is larger than the set lower limit value Dl and smaller than the upper limit value Dh (Dl <Hd <Dh), it is determined as Yes, otherwise it is determined as No. Step S32 is an inspection item for inspecting the extracted pattern Pa in terms of the deviation Hd of the luminance histogram in the color inspection. The lower limit value Dl and the upper limit value Dh can be changed as appropriate. For this reason, the mode of luminance distribution of each pixel in the image data of the extraction pattern Pa can be appropriately set according to the variation in the pattern (inspection target pattern Po) of the inspection target (semiconductor element 16 in this embodiment). .

  In step S33, following the determination that the deviation Hd in step S32 is within the allowable range, the rough evaluation value Vr of the extraction pattern Pa is calculated, and the process proceeds to step S34. In this step S33, the image data of the extraction pattern Pa is differentiated, and the sum of the differential values is divided by the number of pixels of the extraction pattern Pa to calculate the rough evaluation value Vr of the extraction pattern Pa. In this embodiment, in step S33, the rough evaluation value Vr is calculated by handling the extraction pattern Pa as RGB image data.

  In step S34, following the calculation of the rough evaluation value Vr of the extraction pattern Pa in step S33, it is determined whether or not the rough evaluation value Vr is within the allowable range. If Yes, the process proceeds to step S35. Advances to step S36. In this step S34, if the rough evaluation value Vr is larger than the set lower limit value Rl and smaller than the upper limit value Rh (Rl <Vr <Rh), it is determined as Yes, otherwise it is determined as No. To do. Step S34 is an inspection item for inspecting the extracted pattern Pa from the viewpoint of the rough evaluation value Vr in the color inspection. The lower limit value Rl and the upper limit value Rh can be changed as appropriate. Therefore, the degree of allowable color change in the image data of the extraction pattern Pa can be set as appropriate according to the variation in the pattern (inspection target pattern Po) of the inspection target (semiconductor element 16 in this embodiment).

  In step S35, following the determination that the rough evaluation value Vr in step S34 is within the allowable range, it is assumed that all the color inspections of the inspection target pattern Po are completed and the criteria of each item are satisfied. finish. In this step S35, it is assumed that there is no problem in the color inspection of the inspection target pattern Po of the semiconductor element 16 as the inspection target at the observation position, and in order to perform the boundary inspection processing of the inspection target pattern Po in the analysis processing, The taste inspection process is terminated. At this time, the fact that there was no problem in the color inspection is stored in combination with the semiconductor element 16 to be inspected.

  In step S36, following the determination that the deviation Hd in step S32 is not within the allowable range, or the determination that the rough evaluation value Vr is not in the allowable range in step S34, the color inspection of the inspection target pattern Po. This flowchart is terminated on the assumption that it did not complete because it did not satisfy the criteria of any item. In this step S36, the color inspection process is terminated because there is a problem in the color inspection of the inspection target pattern Po of the semiconductor element 16 as the inspection target at the observation position. In this embodiment, when the color inspection process (flowchart in FIG. 10) is finished through step S36, the inspection target pattern Po of the semiconductor element 16 as the inspection target at the observation position is problematic. The analysis process for the semiconductor element 16 is terminated without performing the above inspection. At this time, the fact that there is a problem in any item is stored in combination with the semiconductor element 16 to be inspected. Thereby, the color inspection in the extraction pattern Pa (its image data) as the inspection target pattern Po of the semiconductor element 16 (inspection target) at the observation position is completed.

  Thereafter, the image processing unit 32 performs a boundary inspection process for the inspection target pattern Po in the analysis process. In this boundary inspection process, it is determined whether or not the boundary position and shape in the extraction pattern Pa (image data) described above are correct. FIG. 12 is a flowchart illustrating an example of the content of the boundary inspection process of the inspection target pattern Po in the image processing unit 32. FIG. 13 is an explanatory diagram showing the boundary coordinate direction Y and the scanning state in the partitioned area Dn, where (a) shows the state in the partitioned area D2, and (b) shows the state in the partitioned area D5. Yes. FIG. 14 is an explanatory diagram showing the boundary line (outer contour) of the sample pattern Pm as the origin in the boundary coordinate direction Y in order to explain how the average boundary image Ib of the extraction pattern Pa is generated taking the partition region D2 as an example. (A) schematically shows the boundary line (its image data) of the extracted pattern Pa superimposed on the boundary line of the sample pattern Pm, and (b) schematically shows the average boundary image Ib. FIG. 15 is an explanatory diagram showing a reference boundary image Bb generated from the average boundary image Ib with the boundary line (outer contour) of the sample pattern Pm as the origin in the boundary coordinate direction Y. FIG. 16 is an explanatory diagram showing how the defect candidate Lc is extracted by obtaining the difference between the reference boundary image Bb (its image data) and the extraction pattern Pa (its image data). In FIGS. 14 to 16, for easy understanding, the unevenness in the boundary coordinate direction Y on the boundary line (outer) of the extracted pattern Pa is highlighted, and it does not necessarily match the actual extracted pattern Pa. Not what you want. Hereinafter, each step of the flowchart of FIG. 12 which is an example of the control process in the image processing unit 32 in the boundary inspection process will be described with reference to FIGS. 13 to 16.

  In step S41, an average boundary image Ib in each partitioned area Dn of the extraction pattern Pa generated in the flowchart of FIG. 6 is generated, and the process proceeds to step S42. In step S41, an average boundary image Ib is generated for each partition area Dn (see FIG. 4) set by the sample pattern Pm in the image data of the extraction pattern Pa. The average boundary image Ib is defined as a boundary coordinate direction Y (see FIG. 13) in each partition region Dn with a direction orthogonal to the boundary line (outer) in the sample pattern Pm corresponding to each partition region Dn. The average value of the boundary line (outline) seen in the boundary coordinate direction Y in each partition area Dn is imaged as a unit length line segment. In order to calculate the average boundary image Ib, in this embodiment, as shown in FIG. 13, the scanning in the direction along the boundary line (outer) in the sample pattern Pm corresponding to each partition region Dn is performed in the boundary coordinate direction. This is performed with a predetermined width centered on the boundary line as seen by Y. In this scanning, the extraction pattern Pa is handled as RGB image data, and the brightness at each point on each scanning line is acquired for each color (RGB). For each of the scanning lines, the brightness at each point is integrated, and the integrated value is divided by the obtained number of points to calculate the average value of the brightness for each color (RGB) on each scanning line. Can do. Here, there is a color difference at the boundary line (outside) of the extraction pattern Pa, and the color difference can be detected by a brightness difference for each color (RGB). For this reason, the average value of the brightness for each color (RGB) in each scanning line, that is, the average value of the brightness for each color (RGB) viewed in the boundary coordinate direction Y is reproduced as an image, thereby obtaining the average boundary image Ib. (See FIG. 14B). In the example shown in FIG. 14, the average boundary image Ib is generated as shown in (b) by scanning the boundary line (outside) of the extraction pattern Pa of the partitioned area D2 shown in (a).

  In step S42, following the generation of the average boundary image Ib in each partitioned region Dn in step S41, it is determined whether the matching rate Pc between the average boundary image Ib and the sample pattern Pm is within an allowable range, If yes, go to Step S43, if No, go to Step S47. In this step S42, if the matching rate Pc between the average boundary image Ib and the sample pattern Pm is larger than the allowable value Tp set as a predetermined value (Pc> Tp), it is determined as Yes, and if it is smaller, it is determined as No. Step S42 is an inspection item for inspecting the extracted pattern Pa from the viewpoint of the matching rate Pc in the boundary inspection. This allowable value Tp can be changed as appropriate. For this reason, the difference of the average boundary image Ib with respect to the sample pattern Pm can be appropriately set according to the variation in the pattern (inspection target pattern Po) of the inspection target (semiconductor element 16 in this embodiment).

  In step S43, following the determination that the matching rate Pc between the average boundary image Ib and the sample pattern Pm in step S42 is within the allowable range, a reference boundary image Bb in each partition region Dn is generated, and the process proceeds to step S44. move on. In this step S43, in each divided area Dn, a reference boundary image Bb used as a reference for comparison with the extracted pattern Pa is generated for extracting a defect candidate Lc to be described later in the extracted pattern Pa. In the present embodiment, the reference boundary image Bb is generated by expanding the average boundary image Ib generated in each partition region Dn in the length direction so as to be adapted to the length dimension of the partition region Dn. This length direction is the direction in which the boundary line of the sample pattern Pm in the partition region Dn extends, that is, the direction orthogonal to the boundary coordinate direction Y, and is the scanning direction in step S41. In the example shown in FIG. 15, the average boundary image Ib (see FIG. 14B) corresponding to the partition area D2 is expanded in the length direction so as to match the length dimension of the partition area D2, and in the partition area D2. A reference boundary image Bb is generated.

  In step S44, following the generation of the reference boundary image Bb in each partition area Dn in step S43, the defect candidate Lc is extracted, and the process proceeds to step S45. In this step S44, a difference (see reference numeral Gi shown in FIG. 16) between the reference boundary image Bb (its image data) and the extraction pattern Pa (its image data) in each partition area Dn is obtained (see FIGS. 15 and 16). reference). Thereafter, when there is a location exceeding the allowable value σ with reference to the boundary line (outline) in each partition area Dn in the sample pattern Pm when viewed in the boundary coordinate direction Y, the position is a defect candidate in each partition area Dn. Extracted as Lc (see FIG. 16). The defect candidate Lc is handled as a region surrounded by a line segment indicating the allowable value σ and a line segment indicating the difference Gi. In the example shown in FIG. 16, one defect candidate Lc is extracted in the partitioned area D2 of the extraction pattern Pa.

  In step S45, following the extraction of the defect candidate Lc in each partition area Dn in step S44, it is determined whether or not the defect candidate Lc is within the allowable range. If Yes, the process proceeds to step S46, and if No. Advances to step S47. In step S45, it is determined whether or not the extracted defect candidate Lc is to be detected as a defect by determining whether or not the extracted defect candidate Lc is smaller than the set size. In the present embodiment, as the set size, a size dimension (width dimension) viewed in one direction, a size dimension (length dimension) viewed in the other direction orthogonal to one direction, and an area, Matters are set. Therefore, in step S45, when the width dimension in the defect candidate Lc is smaller than the set size, the length dimension in the defect candidate Lc is smaller than the set size, and the area in the defect candidate Lc is smaller than the set size, It is determined as Yes, and otherwise determined as No. Step S45 is an inspection item for inspecting the extracted pattern Pa from the viewpoint of the size of the defect candidate Lc in the boundary inspection. This set size (width dimension, length dimension and area) can be changed as appropriate. Further, the items in the set size may be set as appropriate and are not limited to the present embodiment. For this reason, the size of the allowable defect candidate Lc can be appropriately set according to the variation in the pattern (inspection target pattern Po) of the inspection target (semiconductor element 16 in this embodiment).

  In step S46, following the determination that the defect candidate Lc is smaller than the set size in step S45, all the boundary inspections of the inspection target pattern Po are completed, and the process is completed assuming that the criteria of each item are satisfied. . In this step S46, the boundary inspection process is terminated because there is no problem in the boundary inspection of the inspection target pattern Po of the semiconductor element 16 as the inspection target at the observation position. At this time, the fact that there was no problem in the boundary inspection is stored in combination with the semiconductor element 16 to be inspected.

  In step S47, it is determined that the matching rate Pc between the average boundary image Ib and the sample pattern Pm in step S42 is not within the allowable range, or that the defect candidate Lc in step S45 is not within the allowable range. Subsequently, it is assumed that the boundary inspection of the inspection target pattern Po has not been completed because the criterion of any item is not satisfied, and this flowchart is ended. In this step S47, the boundary inspection process is terminated assuming that there is a problem in the boundary inspection of the inspection target pattern Po of the semiconductor element 16 as the inspection target at the observation position. At this time, the fact that there is a problem in any item is stored in combination with the semiconductor element 16 to be inspected. Thereby, the boundary inspection in the extraction pattern Pa (its image data) as the inspection target pattern Po of the semiconductor element 16 (inspection target) at the observation position is completed.

  As described above, the appearance inspection apparatus 10 creates a chip map for the wafer 15 to be inspected and held by the operation holding mechanism 13, and determines an observation position by the observation optical system based on the chip map. An analysis process is performed on the inspection object (semiconductor element 16) at the observation position by the image processing unit 32, and the quality of the inspection object (semiconductor element 16) is determined. In the analysis processing in the image processing unit 32, an extraction pattern Pa (image data thereof) is generated in the extraction processing (flowchart of FIG. 6), and a shape inspection (flowchart of FIG. 9) is executed on the extraction pattern Pa. If there is no problem, a color inspection (flowchart in FIG. 10) is performed on the extracted pattern Pa. If there is no problem, a boundary inspection (flowchart in FIG. 12) is performed on the extracted pattern Pa. If not, it is determined that the inspection target (semiconductor element 16 (its inspection target pattern Po)) is a non-defective product. Further, in the analysis processing of the image processing unit 32, if there is any problem in each item of each inspection (shape, color, and boundary), the inspection object (semiconductor element 16 (its inspection object pattern Po)) is a non-defective product. Judge that is not. As a result, the appearance inspection apparatus 10 can inspect all inspection objects (semiconductor elements 16 (its inspection object pattern Po)) on the wafer 15 sucked and held by the operation holding mechanism 13.

  In the appearance inspection apparatus 10 according to the present invention, in the boundary inspection process of the inspection target pattern Po of the analysis process in the image processing unit 32, as shown in FIG. 12, the extracted pattern Pa (its image data) is extracted from each partitioned area Dn. An item (step S42) for determining whether the matching rate Pc between the generated average boundary image Ib and the sample pattern Pm is within the allowable range, and an item for determining whether the defect candidate Lc is within the allowable range. (Step S45) is executed and is completed, and it is determined that there is no problem in the boundary inspection unless there is a large deviation and there is no large unevenness (defect) in the partitioned region Dn unit. The sample pattern Pm has a different shape from the boundary line (outer), or the boundary line (outer) is formed at a position different from the sample pattern Pm. Even if the have irregularities (defects) or are formed, it can be judged as a good product.

  Further, in the appearance inspection apparatus 10, in the boundary inspection process, whether the matching rate Pc between the average boundary image Ib generated from the extracted pattern Pa (image data thereof) of each partition area Dn and the sample pattern Pm is within an allowable range. Are completed by executing the item for determining whether or not the defect candidate Lc is within the allowable range (step S45), and appropriately setting each allowable range. Therefore, it is possible to appropriately set the size of the problem in the deviation of the boundary line (outer) with respect to the sample pattern Pm and the size of the problem in the unevenness (defect) not in the sample pattern Pm. If there is a large deviation or a large unevenness (defect) when viewed in Dn units, it is judged that there is a problem in the boundary inspection. Can.

  Furthermore, in the appearance inspection apparatus 10, in the analysis processing in the image processing unit 32, in the extraction processing of the inspection target pattern Po of the inspection target (semiconductor element 16), as shown in FIG. In order to generate (data), a pixel close to the color centroid in the sample pattern Pm is extracted from each pixel in the search area Ar (step S2), and the search area Ar includes the sample pattern Pm and includes the sample pattern Pm. Since the inspection object pattern Po of the inspection object (semiconductor element 16) has a shape that protrudes from the sample pattern Pm, the inspection object (semiconductor element 16) is a target for each inspection (shape, color, and boundary). be able to.

  In the appearance inspection apparatus 10, since the amount of expansion in the search area Ar, that is, the size of the search area Ar can be changed as appropriate, the range to be subjected to each inspection (shape, color and boundary) is appropriately set. Therefore, inspection accuracy can be ensured.

  In the appearance inspection apparatus 10, during the analysis processing in the image processing unit 32, in the extraction processing of the inspection target pattern Po of the inspection target (semiconductor element 16), as shown in FIG. The extracted pattern Pa is generated by applying the image data as the sample pattern Pm corresponding to the confirmed area Af to the inside of the confirmed area Af applied to the image data) (step S4), and the confirmed area Af Since the amount of contraction, that is, the size of the defined region Af can be changed as appropriate, an extraction pattern Pa (see FIG. 8) in which the hollow portion E (see FIGS. 3 and 7) is filled is generated. The extraction pattern Pa representing only the outer shape of the inspection target pattern Po required for each inspection (shape, color and boundary) can be generated. For this reason, each test | inspection (a shape, a color taste, and a boundary) can be performed more simply and appropriately.

  In the appearance inspection apparatus 10, in the shape inspection process of the inspection target pattern Po of the analysis process in the image processing unit 32, as shown in FIG. 9, an item for determining whether or not the total area Sa is within the allowable range (step) S12), an item for determining whether or not the peripheral length La is within the allowable range (step S14), an item for determining whether or not the vertical dimension Mla is within the allowable range (step S16), and a horizontal dimension An item for determining whether or not Mwa is within the allowable range (step S18) and an item for determining whether or not the correlation value C is within the allowable range (step S20) are executed and completed. In addition, since each allowable range can be set as appropriate, it is determined that there is no problem if they are similar to some extent. Therefore, the shape may be different from the sample pattern Pm or may be different from the sample pattern Pm. Even if it is formed at a position different from the sample pattern Pm or has irregularities (defects) that are not in the sample pattern Pm, it can be determined as non-defective and the inspection accuracy is high. Can be secured.

  In the appearance inspection apparatus 10, in the color inspection process of the inspection target pattern Po of the analysis process in the image processing unit 32, as shown in FIG. 10, is the deviation Hd of the luminance histogram of the extracted pattern Pa within an allowable range? An item for determining whether or not (step S32) and an item for determining whether or not the rough evaluation value Vr is within the allowable range (step S34). Since it can be set as appropriate, it is judged that there is no problem unless there is a portion where the overall color tone is similar to some extent and the color changes extremely. Therefore, the color may be different from the sample pattern Pm. Even if the color scheme is different from Pm, it can be determined as a non-defective product and the inspection accuracy can be ensured.

  The appearance inspection apparatus 10 includes an item (step S34) for determining whether or not the rough evaluation value Vr is within an allowable range in the color inspection process of the inspection target pattern Po of the analysis process in the image processing unit 32. Therefore, even if the deviation Hd of the histogram of the luminance of the extracted pattern Pa as a whole is within the allowable range, it is judged that there is a problem in the color inspection if there is a portion where the color changes extremely. Inspection accuracy can be ensured. For example, as shown in FIG. 17A, the extracted pattern Pa (inspection target pattern Po) is in a gradation shape even when the deviation Hd of the luminance histogram of the extracted pattern Pa as a whole is within the allowable range. In some cases, there is no performance problem, but as shown in FIG. 17B, when the extraction pattern Pa (inspection target pattern Po) is patchy, there is a performance problem. This is because it is necessary to determine that only the pattern Pa (inspection target pattern Po) has a problem.

  Therefore, in the appearance inspection apparatus 10 according to the present invention, in the range where there is no problem in performance, the formed position and the color unevenness are different, and the difference is not observed in a certain tendency (there is variation). Inspection accuracy can be secured even for the target pattern Po.

  In the above-described embodiments (including modifications), the appearance inspection apparatus 10 as an example of the appearance inspection apparatus according to the present invention has been described, but an observation mechanism that acquires an image of an inspection target pattern of an inspection target; An image inspection unit comprising: an image processing unit that analyzes image data acquired by the observation mechanism using a sample pattern prepared in advance and determines the quality of the inspection object, wherein the image processing unit For each of a plurality of partition areas along the boundary line of the sample pattern, an average boundary image that is an average value of the boundary line of the inspection target pattern viewed in the boundary coordinate direction orthogonal to the boundary line of the sample pattern is generated. And an item for determining whether or not the matching ratio between the average boundary image and the sample pattern is within an allowable range, a reference boundary image generated from the average boundary image, and the inspection target pattern. It is only necessary to extract a defect candidate by obtaining a difference, and determine whether or not the defect candidate is within an allowable range, and an appearance inspection apparatus that performs a boundary inspection process of the inspection target pattern. The present invention is not limited to the above-described embodiments.

  In the above-described embodiment, when the average boundary image Ib is generated (step S41) in the boundary inspection process (flowchart in FIG. 12), the scanning is performed in the direction along the boundary line (outline) in the sample pattern Pm. However, if it is possible to generate the average boundary image Ib from the average value of the brightness of each color (RGB) viewed in the boundary coordinate direction Y, for example, scanning is performed in the boundary coordinate direction Y. The search range may be matched with the length dimension of the boundary line (outer) in the sample pattern Pm in each partition area Dn, and is not limited to the above-described embodiment.

  Furthermore, in the above-described embodiment, after the extraction pattern Pa (its image data) is generated, shape inspection (the flowchart of FIG. 9) is performed on the extraction pattern Pa. The color inspection (flowchart in FIG. 10) is executed, and if there is no problem, the boundary inspection (flowchart in FIG. 12) is executed on the extracted pattern Pa. And the order of executing the boundary) may be set as appropriate, and is not limited to the above-described embodiments.

  In the above-described embodiment, the item (step S12) for determining whether or not the total area Sa is within the allowable range is performed during the shape inspection (the flowchart of FIG. 9), and the peripheral length La is within the allowable range thereafter. Whether or not the vertical dimension Mla is within the permissible range (step S16), and then the horizontal dimension Mwa is within the permissible range (step S16). The item (step S18) is determined, and then the item (step S20) is determined to determine whether or not the correlation value C is within the allowable range. What is necessary is just to set and it is not limited to the above-mentioned Example.

  In the above-described embodiment, the item (step S32) for determining whether or not the deviation Hd of the luminance histogram of the extracted pattern Pa is within the allowable range at the time of the color inspection (flowchart in FIG. 10) is performed and then roughened. Although the item for determining whether or not the evaluation value Vr is within the allowable range (step S34) is performed, the order in which each item is executed may be set as appropriate, and is limited to the above-described embodiment. It is not a thing.

  In the above-described embodiment, the matching rate Pc between the average boundary image Ib generated from the extracted pattern Pa (its image data) and the sample pattern Pm is within the allowable range in each partition area Dn during the boundary inspection (flowchart in FIG. 12). The item (step S42) for determining whether or not the defect candidate Lc is within the allowable range is performed, and then the item (step S45) for determining whether or not the defect candidate Lc is within the allowable range is performed. The order to be performed may be set as appropriate, and is not limited to the above-described embodiment.

  As described above, the appearance inspection apparatus according to the present invention has been described based on the embodiments. However, the specific configuration is not limited to the embodiments, and design changes and additions are allowed without departing from the gist of the present invention. Is done.

DESCRIPTION OF SYMBOLS 10 Appearance inspection apparatus 11 Observation mechanism 16 Semiconductor element (As an example of inspection object) 32 Image processing part Af Determining area Ar Search area Bb Reference | standard boundary image C Correlation value Dn Partition area Hd (Difference of the brightness | luminance histogram of an extraction pattern) Ib Average boundary image La Perimeter length Lc Defect candidate Lm Perimeter length Mla Vertical dimension Mwa Horizontal dimension Pa Extraction pattern Pc Matching rate Pm Sample pattern Po Inspection pattern Sa Total area Vr Roughness evaluation value Y Boundary coordinate direction

Claims (4)

An observation mechanism that acquires an image of an inspection target pattern of the inspection target, and an image processing unit that analyzes image data acquired by the observation mechanism using a sample pattern prepared in advance and determines whether the inspection target is good or bad. A visual inspection device comprising:
The image processing unit, as analysis processing,
For each of a plurality of partition regions along the boundary line of the sample pattern, generate an average boundary image that is an average value of the boundary line of the inspection target pattern viewed in a boundary coordinate direction orthogonal to the boundary line of the sample pattern; An item for determining whether the matching ratio between the average boundary image and the sample pattern is within an allowable range;
An item for extracting a defect candidate by obtaining a difference between the reference boundary image generated from the average boundary image and the inspection target pattern, and determining whether the defect candidate is within an allowable range;
A visual inspection apparatus that performs a boundary inspection process on the inspection target pattern.   The image processing unit extracts pixels close to the color centroid of the sample pattern from among the pixels in the search region set to a region larger than the sample pattern while including the sample pattern, and the pixel group from which the color is extracted The fixed area set as an area smaller than the sample pattern is applied to the inside of the sample pattern, and the image data as the sample pattern at the position corresponding to the fixed area is applied to the inside of the fixed area and extracted. The appearance inspection apparatus according to claim 1, wherein a pattern is generated and the extracted pattern is used for an analysis process of the inspection target pattern. The image processing unit, as analysis processing,
An item for determining whether the total area of the extraction pattern is within an allowable range;
An item for determining whether the perimeter of the extraction pattern is within an allowable range;
An item for determining whether the vertical dimension of the extraction pattern is within an allowable range;
An item for determining whether or not a lateral dimension of the extraction pattern is within an allowable range;
An item for determining whether or not a correlation value indicating a degree of difference between the extracted pattern and the sample pattern is within an allowable range;
The appearance inspection apparatus according to claim 2, wherein a shape inspection process of the inspection target pattern is performed. The image processing unit, as analysis processing,
An item for determining whether or not the deviation of the luminance histogram of the extracted pattern is within an allowable range;
Items for determining whether or not the rough evaluation value calculated by differentiating the image data of the extracted pattern and dividing the sum of the differential values by the number of pixels of the extracted pattern is within an allowable range;
The appearance inspection apparatus according to claim 2, wherein a color inspection process of the inspection target pattern is performed.